In automated form processing, the handwritten documents, which may include handwritten content, are scanned to generate electronic documents. Subsequently, the electronic documents may be processed to extract information included in those electronic documents. Typically, the information in the handwritten documents is included in different regions of interest (e.g., fields of an application form) and the electronic documents may be processed to extract such regions of interest using pre-defined templates of the electronic documents. Thereafter, the information from the extracted regions of interest may be determined using known techniques, such as optical character recognition (OCR), intelligent character recognition (ICR), and the like. In a similar way, electronic documents including printed information (e.g., application number and barcode) may be processed to determine the information included in regions of interests.
However, in certain scenarios, the users may write outside the specified regions of interest. Further, it might be possible that during scanning of the handwritten documents, skew is introduced in the electronic documents (e.g., due to deformation of the handwritten documents). In such scenarios, extracting the information, using the pre-defined templates may lead to incorrect extraction of regions of interests, and subsequent incorrect recognition of the information.